Quantitative Precipitation Estimation with Ground-Based Phased Array Radars

Radar observations are key to improving the physical understanding of atmospheric processes, such as the coupled process continuum of clouds, convection and precipitation, a priority “designated observable” identified in the 2017 Earth Science Decadal Survey. Because atmospheric microphysical observations associated with clouds, convection, and precipitation are constrained by the capabilities of the instruments used to measure them, there is a need to understand the information content that can be extracted from active sensors to optimize the usage of current networks and satellite constellations, the design of future instruments, and their synergistic combination. This work takes a step in this direction with a simulation framework and first observations collected by a ground-based phased-array radar (PAR). It explores explainable machine learning (ML) to develop a backward operator that predicts probabilistic microphysical characteristics and precipitation rates from radar returns to account for uncertainty and errors.